Managing Pathogen-related Health Risks for Different Water Types with Bayesian Networks

Monitoring and predicting health-related water quality is critical for water utilities managing a range of water types, from drinking water to wastewater. Traditional quantitative microbial risk assessment, as well as pathogen measurement methods, are time-consuming and resource-intensive. To complement these approaches in cases where several scenarios or predictions must be run in a limited amount of time, a number of Bayesian Networks were developed to predict pathogen counts or related health risks for (1) a drinking water source; (2) untreated stormwater, (3) a drinking water source augmented with treated stormwater, (4) recreational waters and (5) different wastewater reuse options. The study locations were situated in Australia, and the models were trained using empirical data, outputs from previous studies available from the literature, and experts’ input. By applying a participatory modelling approach, the conceptual models were developed through stakeholder workshops, which were also used to validate the final models and to collect qualitative data when the empirical, numerical data was insufficient. The advantages of Bayesian Networks include the ability of dealing with uncertainty and missing data, which are commonplace for this kind of modelling and measurement applications, as well as the quickness of the simulations, which allowed to obtain predictions in near real-time or to evaluate several different scenarios in a matter of minutes. In addition, the involvement of relevant stakeholders during the course of the whole model development process, ensured that the model structure and outputs were clear and agreed by all the end users; this in turn increased the confidence in the models and maximised their deployment. As a result, the developed Bayesian Networks were effectively used, among others, for (1) climate change impacts estimation, (2) water resource management optimisation, and (3) real-time management of major swimming events; proving to be, under certain conditions, a cost-effective, alternative pathogen-related risk assessment and management modelling approach for water utilities.